[Example] 390 — Loom Video Transcription with Deepgram STT (Node.js)

examples/390-loom-transcription-node/.env.example

@@ -0,0 +1,5 @@
+# Deepgram — https://console.deepgram.com/
+DEEPGRAM_API_KEY=
+
+# Loom — https://www.loom.com/developer-portal
+LOOM_API_KEY=

examples/390-loom-transcription-node/.gitignore

@@ -0,0 +1 @@
+node_modules/

examples/390-loom-transcription-node/README.md

@@ -0,0 +1,60 @@
+# Loom Video Transcription with Deepgram
+
+Transcribe Loom video recordings using Deepgram's nova-3 pre-recorded speech-to-text. This example builds a Node.js Express server that accepts a Loom share URL, fetches the video via the Loom Developer API, and returns a full transcript powered by Deepgram.
+
+## What you'll build
+
+A Node.js server that takes a Loom video URL, downloads the recording from Loom's CDN via their Developer API, and sends it to Deepgram pre-recorded STT for a complete transcript with speaker diarization and smart formatting. Useful for teams who use Loom for async communication and want searchable, indexed transcripts.
+
+## Prerequisites
+
+- Node.js 18+
+- Deepgram account — [get a free API key](https://console.deepgram.com/)
+- Loom Developer account — [access the developer portal](https://www.loom.com/developer-portal)
+
+## Environment variables
+
+| Variable | Where to find it |
+|----------|-----------------|
+| `DEEPGRAM_API_KEY` | [Deepgram console](https://console.deepgram.com/) |
+| `LOOM_API_KEY` | [Loom developer portal](https://www.loom.com/developer-portal) → API Keys |
+
+## Install and run
+
+```bash
+cp .env.example .env
+# Add your API keys to .env
+
+npm install
+npm start
+```
+
+Then send a Loom video for transcription:
+
+```bash
+curl -X POST http://localhost:3000/transcribe \
+  -H "Content-Type: application/json" \
+  -d '{"url": "https://www.loom.com/share/YOUR_VIDEO_ID"}'
+```
+
+## Key parameters
+
+| Parameter | Value | Description |
+|-----------|-------|-------------|
+| `model` | `nova-3` | Deepgram's most accurate general-purpose model |
+| `smart_format` | `true` | Adds punctuation, capitalisation, and number formatting |
+| `diarize` | `true` | Labels each word with a speaker number for multi-speaker recordings |
+| `paragraphs` | `true` | Groups transcript into readable paragraphs |
+
+## How it works
+
+1. Client sends a Loom share URL to `POST /transcribe`
+2. Server extracts the video ID from the URL
+3. Server calls the Loom Developer API to get video metadata and a time-limited download URL
+4. Server downloads the video file from Loom's CDN
+5. Server sends the video buffer to Deepgram's pre-recorded STT API (Deepgram accepts video files directly — no audio extraction needed)
+6. Server returns the transcript with word count, duration, and speaker labels
+
+## Starter templates
+
+[deepgram-starters](https://github.com/orgs/deepgram-starters/repositories)